Training error about rasa HOT 6 CLOSED

Hi @yogeshhk - we're using scikit-learn 0.18.1 - can you tell me exactly which line throws the importerror?

from rasa.

Traceback (most recent call last):
File "C:\Users\kulkarni\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.6.2.3262.win-x86_64\lib\runpy.py", line 162, in _run_module_as_main
"main", fname, loader, pkg_name)
File "C:\Users\kulkarni\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.6.2.3262.win-x86_64\lib\runpy.py", line 72, in _run_code
exec code in run_globals
File "build\bdist.win-amd64\egg\rasa_nlu\train.py", line 65, in
File "build\bdist.win-amd64\egg\rasa_nlu\train.py", line 54, in do_train
File "build\bdist.win-amd64\egg\rasa_nlu\train.py", line 31, in create_trainer

File "build\bdist.win-amd64\egg\rasa_nlu\trainers\spacy_sklearn_trainer.py", line 8, in
File "build\bdist.win-amd64\egg\rasa_nlu\classifiers\sklearn_intent_classifier.py", line 1, in
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\model_selection_init_.py", line 1, in
from .split import BaseCrossValidator
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\model_selection_split.py", line 36, in
from ..gaussian_process.kernels import Kernel as GPKernel
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\gaussian_process_init.py", line 13, in
from .gpr import GaussianProcessRegressor
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\gaussian_process\gpr.py", line 15, in
from sklearn.gaussian_process.kernels import RBF, ConstantKernel as C
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\gaussian_process\kernels.py", line 30, in
from ..metrics.pairwise import pairwise_kernels
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\metrics_init_.py", line 33, in
from . import cluster
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\metrics\cluster_init_.py", line 23, in
from .bicluster import consensus_score
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\metrics\cluster\bicluster_init_.py", line 1, in
from .bicluster_metrics import consensus_score
File "c:\users\kulkarni\appdata\local\enthought\canopy\user\lib\site-packages\sklearn\metrics\cluster\bicluster\bicluster_metrics.py", line 6, in
from sklearn.utils.validation import check_arrays
ImportError: cannot import name check_arrays

from rasa.

my best guess is there is something up with your installation. I'm not familiar with the canopy distribution, but my advice would be to do a fresh install, with scikit-learn 0.18.1.
Then open up a python interpreter, and run
from sklearn import model_selection
and see if that throws the same error.

from rasa.

Thanks for the suggestion. I removed the Canopy distribution. Installed Minicoda 2 and then compilation seems to be fine. I could train the model and added it to the config.json I ran the server by

python -m rasa_nlu.server -c config.json

Then fired the test command from another window:

curl -XPOST localhost:5000/parse -d '{"q":"I am looking for Chinese food"}' | python -mjson.tool

It said:

% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 35 0 0 100 35 0 149 --:--:-- --:--:-- --:--:-- 149
No JSON object could be decoded

I tried another test. In browser I pasted:

http://localhost:5000/parse?q=I am looking for Chinese food

It gave:

{"text": "I am looking for Chinese food", "intent": "restaurant_search", "entities": []}

I did not get the correct reply as:

{
  "intent" : "restaurant_search",
  "entities" : [
    {
      "start": 8,
      "end": 15,
      "value": "chinese",
      "entity": "cuisine"
    }
  ]
}

What could have gone wrong?

There was exception on server side. Traceback is:

Exception happened during processing of request from ('127.0.0.1', 49833)
Traceback (most recent call last):
File "C:\Users\kulkarni\Miniconda2\lib\SocketServer.py", line 290, in handle_request_noblock
self.process_request(request, client_address)
File "C:\Users\kulkarni\Miniconda2\lib\SocketServer.py", line 318, in process_request
self.finish_request(request, client_address)
File "C:\Users\kulkarni\Miniconda2\lib\SocketServer.py", line 331, in finish_request
self.RequestHandlerClass(request, client_address, self)
File "build\bdist.win-amd64\egg\rasa_nlu\server.py", line 76, in
File "build\bdist.win-amd64\egg\rasa_nlu\server.py", line 160, in init
File "C:\Users\kulkarni\Miniconda2\lib\SocketServer.py", line 652, in init
self.handle()
File "C:\Users\kulkarni\Miniconda2\lib\BaseHTTPServer.py", line 340, in handle
self.handle_one_request()
File "C:\Users\kulkarni\Miniconda2\lib\BaseHTTPServer.py", line 328, in handle_one_request
method()
File "build\bdist.win-amd64\egg\rasa_nlu\server.py", line 201, in do_POST
File "C:\Users\kulkarni\Miniconda2\lib\json_init.py", line 339, in loads
return _default_decoder.decode(s)
File "C:\Users\kulkarni\Miniconda2\lib\json\decoder.py", line 364, in decode
obj, end = self.raw_decode(s, idx=_w(s, 0).end())
File "C:\Users\kulkarni\Miniconda2\lib\json\decoder.py", line 382, in raw_decode
raise ValueError("No JSON object could be decoded")
ValueError: No JSON object could be decoded

My Model training_data.json does have entry like:

      {
        "text": "show me chinese restaurants", 
        "intent": "restaurant_search", 
        "entities": [
          {
            "start": 8, 
            "end": 15, 
            "value": "chinese", 
            "entity": "cuisine"
          }
        ]
      }, 

I thought it would pick up entities, but did not. Any clues?

from rasa.

The example data provided is really just a toy example to show how to use rasa, & doesn't provide enough examples for the spacy NER model to pick up chinese as an entity. I'll add a comment about that to the docs :)

Regarding the issue with your curl POST, my best guess is that the windows version of cURL doesn't automatically encode the json data for you. I would suggest doing the request with Postman or a similar API testing tool (or just the python requests package

from rasa.

closing this now. If you are certain the HTTP request is valid & there's an issue on rasa's side, please re-open.

from rasa.